Data Setting Research Articles

Optimal selection of resampling methods for imbalanced data with high complexity.

Class imbalance is a major problem in classification, wherein the decision boundary is easily biased toward the majority class. A data-level solution (resampling) is one possible solution to this problem. However, several studies have shown that resampling methods can deteriorate the classification performance. This is because of the overgeneralization problem, which occurs when samples produced by the oversampling technique that should be represented in the minority class domain are introduced into the majority-class domain. This study shows that the overgeneralization problem is aggravated in complex data settings and introduces two alternate approaches to mitigate it. The first approach involves incorporating a filtering method into oversampling. The second approach is to apply undersampling. The main objective of this study is to provide guidance on selecting optimal resampling methods in imbalanced and complex datasets to improve classification performance. Simulation studies and real data analyses were performed to compare the resampling results in various scenarios with different complexities, imbalances, and sample sizes. In the case of noncomplex datasets, undersampling was found to be optimal. However, in the case of complex datasets, applying a filtering method to delete misallocated examples was optimal. In conclusion, this study can aid researchers in selecting the optimal method for resampling complex datasets.

BUILDING A CONTROL ALGORITHM FOR AUTOMOTIVE AIRBAG RESPONSE CAPABILITY

The article presents the results of developing a control algorithm for the activation time of the driver-side airbag in cars. The authors used C++ programming language to build the control algorithm. The research results show that when applying this algorithm to the airbag control model, some characteristics can be observed and tested, such as the airbag activation time being compatible with the magnitude of impact force. This implies that the larger the impact force (collision), the faster the airbag activation time will be. When the sensor loses its signal, the system will display an error on the car's display screen. This algorithm can also change the input data settings so that users can adjust the data to improve the airbag's optimal operation.

Generalizing supervised deep learning MRI reconstruction to multiple and unseen contrasts using meta-learning hypernetworks

Meta-learning has recently been an emerging data-efficient learning technique for various medical imaging operations and has helped advance contemporary deep learning models. Furthermore, meta-learning enhances the knowledge generalization of the imaging tasks by learning both shared and discriminative weights for various configurations of imaging tasks during training. However, existing meta-learning models attempt to learn a single set of weight initializations of a neural network that might be fundamentally restrictive under the heterogeneous (multimodal) data scenario. This work aims to develop a multimodal meta-learning model for image reconstruction, which augments meta-learning with evolutionary capabilities to encompass diverse acquisition settings of heterogeneous data. Our proposed model called KM-MAML (Kernel Modulation-based Multimodal Meta-Learning), has hypernetworks (auxiliary learners) that evolve to generate mode-specific (or context-specific) weights. These weights provide the mode-specific inductive bias for multiple modes by re-calibrating each kernel of the base network for image reconstruction via a low-rank kernel modulation operation. Furthermore, we incorporate gradient-based meta-learning (GBML) in the contextual space to update the weights of the hypernetworks based on different modes. The hypernetworks and the base reconstruction network in the GBML setting provide discriminative mode-specific features and low-level image features, respectively. We extensively evaluate our model for multi-contrast magnetic resonance image reconstruction considering the essential research directions in fastMRI for multimodal and rich transfer learning capabilities across various MRI contrasts. Our comparative studies show that the proposed model (i) exhibits superior reconstruction performance over joint training, other meta-learning methods, and various context-specific MRI reconstruction architectures, and (ii) better adaptation to 80% and 92% of unseen multi-contrast data contexts with improvement margins of 0.1 to 0.5 dB in PSNR and around 0.01 in SSIM, respectively. Besides, a representation analysis with U-Net as the base network shows that kernel modulation infuses 80% of mode-specific representation changes in the high-resolution layers. Our source code is available at https://github.com/sriprabhar/KM-MAML/.

Application and testing of RFID Software-Based Shielding in real fashion retail stores

PURPOSE: The present paper investigates the real performances of Software-Based Shielding (SBS) in two different stores of fashion retailers located in Northern Italy. DESIGN/METHODOLOGY/APPROACH: The study is based on a double case study analysis. Six different factors have been chosen, with two levels each. Namely, we investigated two different (i) stores; (ii) reader models; (iii) power levels; (iv) classification methods; (v) training data sets and (vi) settings of reference tags. The results have been evaluated in terms of overall and specific accuracies, and in percentage of false front (i.e., tags wrongly located in the sales floor area). FINDINGS: SBS proves to be a sound method for classifying tags’ location during normal operations in real-life stores, with overall accuracy up to 0.95. Of the two readers, reader A shows better results in terms of both overall and front accuracy, whereas reader B performs better in terms of rear accuracy and percentage of false front. With respect to the classification method, the combination of Neural Network with reads from reader A provides the best results. With respect to the training data, the use of front and back reads for training performs mostly better than the training with sole front data. ORIGINALITY: We are not aware of any other study that performed and reported results of SBS testing under normal operations in real stores. To the best of our knowledge, this study is the first one to report such results. RESEARCH LIMITATIONS: Main limitations of our research are the limited set of factors and levels, and the specific Logistic Regression and Neural Network methods that we used. Also, we did not consider tags disposition and density in our study. PRACTICAL IMPLICATIONS: We prove that SBS is a feasible option that could replace physical shielding in retail stores. We call to action for further research on this topic, and for retailers to test it in different store locations.

Behavioural antecedents of Bitcoin trading volume

This paper aims to examine the behavioural determinants of Bitcoin trading volume within a cross-country framework of 14 world economies plus the Eurozone. We introduce a basic taxonomy of behavioural indicators, distinguishing between consumer confidence, economic policy uncertainty (EPU), and indicators of financial volatility. Our estimations reveal that the Bitcoin trading volume can be predicted more accurately by EPU than by any other class of indicators. Finally, we identify the COVID-19 shock as a catalyst for a psychologically-driven Bitcoin market and find evidence that Bitcoin was a macro hedging instrument in the pandemic. To obtain our results, we conducted a panel Granger causality test, employing the Least Squares Dummy Variables (LSDV) estimator. Contrary to previous research, we found that market fundamentals (industrial production and equity market volume) became significant drivers of Bitcoin trading during the pandemic. This conclusion was preserved when we used the LSDV corrected estimator, which is more suitable for panels with a smaller time dimension. Apart from the practical implications for traders, this paper provides researchers with detailed steps for applying Granger causality testing in panel data settings.

Direct Search Techniques for Mixed Stochastic Nonlinear Programming Model

Stochastic programming is a methodology utilized for the purpose of achieving optimal planning and decision-making outcomes when faced with uncertain data. The subject of investigation pertains to a stochastic optimization problem wherein the results of stochastic data are not disclosed during runtime, and the optimization of the decision does not necessitate foresight into forthcoming outcomes. This establishes a strong correlation with the imperative need for immediate optimization in uncertain data settings, enabling effective decision-making in the present moment. The present study introduces a novel methodology for achieving global optimization of the model for nonlinear mixed-stochastic programming problem. The present study centers on stochastic problems that are two-staged and entail non-linearities in both the objective function and constraints. The first stage variables are discrete in nature, whereas the second stage variables are a combination of continuous and mixed types. Scenario-based representations are utilized for formulating problems. The fundamental approach to address the non-linear mixed-stochastic programming problem involves converting the model into a deterministic non-linear mixed-count program that is equivalent in form. The feasibility of this proposition stems from the discrete distribution assumption of uncertainty, which can be represented by a limited set of scenarios. The magnitude of the model size will increase significantly due to the quantity of scenarios and time horizons involved. The utilization of filtered probability space in conjunction with data mining techniques will be employed for the purpose of scenario generation. The methodology employed for addressing nonlinear mixed-integer programming problems of significant scale involves elevating the value of a non-basic variable beyond its boundaries in order to compel a basis variable to attain a cumulative value. Subsequently, the problem is simplified by maintaining a constant count variable and modifying it incrementally in discrete intervals to achieve an optimal solution at a global level.

Evaluating the techno-economic impact of decarbonizing buildings by using borehole heat exchangers in comparison to fuel-based systems

The climate change, volatile geopolitical conditions, energy insecurity, and supply disruptions experienced lately have highlighted the need for a solid strategy to speed up the transition to green energy and reduce dependence on fossil fuels. Developing tools and methods for fast assessment of techno-economic, environmental and socio-economic aspects of sustainable energy solutions will contribute to this goal. This paper presents results of a case study using an inhouse techno-economic analysis tool developed to provide access to detailed data settings, reflecting the most important technical as well as economic parameters of studied energy technologies. Five different energy solutions for heating and cooling of a specified building in a given geographical position and thereby economic environment have been assessed and compared using the inhouse analysis tool. Based on the proposed techno-economic framework, the energy efficiencies, costs, subsidies and negative CO2 emissions are estimated and used to identify the critical economic parameters on profitability and policy making recommendations. Using the coal-based heating system as reference case, the comparative study carried out shows the advantages and disadvantages of biomass and natural gas based boilers as well as the geothermal installation. Considering the environmental benefits of the renewable based energy resources, which will become a stronger market driver for every year. The results show that both air-source and ground source heat pumps are currently profitable under current subsidy schemes. The net present value is highly influenced by capital costs and subsidies. In the same way, natural gas prices significantly affect the net present value of gas boilers. As gas prices increase over 43 % of the current level in Poland, heat pumps become more attractive. Replacing a coal-fired boiler with gas will take up to years to pay-off. However, the air source heat pump and ground source heat pump will take longer to pay off, up to 10 years for the air source heat pump and 15 years for the ground source heat pump. The carbon intensity of the ground source heat pump (186 kg CO2 eq) is lower than the levels of the gas boilers (203 kg CO2 eq) and the air source heat pump (263 kg CO2 eq). Ground source heat pump system installation expects to provide a substantial saving of 58 % on CO2 emissions versus coal-fired systems. Air source heat pump P emits 267 kg CO2/kWh/year more than gas boilers, but emissions decrease drastically as the Polish power grid becomes greener in 2030 and 2040.

Some Clustering Methods, Algorithms and their Applications

Clustering is a type of unsupervised learning [15]. When no target values are known, or "supervisors," in an unsupervised learning task, the purpose is to produce training data from the inputs themselves. Data mining and machine learning would be useless without clustering. If you utilize it to categorize your datasets according to their similarities, you'll be able to predict user behavior more accurately. The purpose of this research is to compare and contrast three widely-used data-clustering methods. Clustering techniques include partitioning, hierarchy, density, grid, and fuzzy clustering. Machine learning, data mining, pattern recognition, image analysis, and bioinformatics are just a few of the many fields where clustering is utilized as an analytical technique. In addition to defining the various algorithms, specialized forms of cluster analysis, linking methods, and please offer a review of the clustering techniques used in the big data setting.

Does Scoring Method Impact Estimation of Significant Individual Changes Assessed by Patient-Reported Outcome Measures? Comparing Classical Test Theory Versus Item Response Theory

ObjectivesThis study aimed to examine the ability of classical test theory (CTT) and item response theory (IRT) scores assessed by Patient-Reported Outcomes Measurement Information System® (PROMIS®) measures to identify significant individual changes in the setting of clinical studies, using both simulated and empirical data. MethodsWe used simulated data to compare the estimation of significant individual changes between CTT and IRT scores across different conditions and a clinical trial data set to verify the simulation results. We calculated reliable change indexes to estimate significant individual changes. ResultsFor small true change, IRT scores showed a slightly higher rate of classifying change groups than CTT scores and were comparable with CTT scores for a shorter test length. Additionally, IRT scores were found to have a prominent advantage in the classification rates of change groups for medium to high true change over CTT scores. Such an advantage became prominent in a longer test length. The empirical data analysis results using an anchor-based approach further supported the above findings that IRT scores can more accurately classify participants into change groups than CTT scores. ConclusionsGiven that IRT scores perform better, or at least comparably, in most conditions, we recommend using IRT scores to estimate significant individual changes and identify responders to treatment. This study provides evidence-based guidance in detecting individual changes based on CTT and IRT scores under various measurement conditions and leads to recommendations for identifying responders to treatment for participants in clinical trials.

Computational Efficient Approximations of the Concordance Probability in a Big Data Setting.

Performance measurement is an essential task once a statistical model is created. The area under the receiving operating characteristics curve (AUC) is the most popular measure for evaluating the quality of a binary classifier. In this case, the AUC is equal to the concordance probability, a frequently used measure to evaluate the discriminatory power of the model. Contrary to AUC, the concordance probability can also be extended to the situation with a continuous response variable. Due to the staggering size of data sets nowadays, determining this discriminatory measure requires a tremendous amount of costly computations and is hence immensely time consuming, certainly in case of a continuous response variable. Therefore, we propose two estimation methods that calculate the concordance probability in a fast and accurate way and that can be applied to both the discrete and continuous setting. Extensive simulation studies show the excellent performance and fast computing times of both estimators. Finally, experiments on two real-life data sets confirm the conclusions of the artificial simulations.

Distributed Computation for Marginal Likelihood based Model Choice

We propose a general method for distributed Bayesian model choice, using the marginal likelihood, where a data set is split in non-overlapping subsets. These subsets are only accessed locally by individual workers and no data is shared between the workers. We approximate the model evidence for the full data set through Monte Carlo sampling from the posterior on every subset generating a model evidence per subset. The results are combined using a novel approach which corrects for the splitting using summary statistics of the generated samples. Our divide-and-conquer approach enables Bayesian model choice in the large data setting, exploiting all available information but limiting communication between workers. We derive theoretical error bounds that quantify the resulting trade-off between computational gain and loss in precision. The embarrassingly parallel nature yields important speed-ups when used on massive data sets as illustrated by our real world experiments. In addition, we show how the suggested approach can be extended to model choice within a reversible jump setting that explores multiple feature combinations within one run.

Integration of sparse and continuous data sets using machine learning for core mineralogy interpretation

In earth science, integrating noninvasive continuous data streams with discrete invasive measurements remains an open challenge. We address such a problem — that of predicting whole-core mineralogy using discrete measurements with the help of machine learning. Our targets are sparsely sampled mineralogy from X-ray diffraction, and features are continually sampled elemental oxides from X-ray fluorescence. Both data sets are acquired on a core cut from a Mississippian-age mixed siliciclastic-carbonate formation in the U.S. midcontinent. The novelty lies in predicting multiple classes of output targets from input features in a small multidimensional data setting. Our workflow has three salient aspects. First, it shows how single-output models are more effective in relating selective target-feature subsets than using a multi-output model for simultaneously relating the entire target-feature set. Specifically, we adopt a competitive ensemble strategy comprising three classes of regression algorithms — elastic net (linear regression), XGBoost (tree-based), and feedforward neural networks (nonlinear regression). Second, it shows that feature selection and engineering, when done using statistical relationships within the data set and domain knowledge, can significantly improve target predictability. Third, it incorporates k-fold cross-validation and grid-search-based parameter tuning to predict targets within 4%–6% accuracy using 40% training data. Results open doors to generating a wealth of information in energy, environmental, and climate sciences where remotely sensed data are inexpensive and abundant but physical sampling may be limited due to analytic, logistic, or economic issues.

Prior authorizations for oral anticancer drugs and delayed prescription fills.

1504 Background: Oral anticancer drugs have proliferated in recent decades. These drugs are increasingly approved based on provisional data via the FDA’s accelerated approval pathway and have high prices, motivating payers to closely monitor how they are used. We assessed use of prior authorization (PA) of oral cancer drugs for Medicare Part D beneficiaries and assessed the impact of PA on delays in prescription fills. Methods: Using Medicare Part D formulary and claims data from 2010-2019, we described trends in Part D plans’ use of PA for oral oncology drugs, identified using the Oncology Care Model drug list (excluding hormonal therapies). We then examined changes in access to imatinib after the introduction of a new PA. We focused on imatinib because it is a highly effective first-in-class drug with a generic entrant in 2016. We compared Part D beneficiaries continuously enrolled in a plan with a policy change (new PA introduced) or a control plan (no change) in the 120 days before and after the policy change or equivalent date (control) and had at least three 30-day imatinib fills in the pre period. This restriction aimed to identify individuals filling regularly and plausibly expected to continue the drug. For each index fill in the post period, we calculated predicted fill (date of previous fill + days supply received). Our outcome variable was the difference (in days) between predicted and observed index fill date, censored at 120 days for post-period observations without an observed fill. We used an event study to model predicted vs. actual time to first fill following a policy change (new PA) relative to the controls (no change), adjusting for patient age, sex, low-income subsidy, average number of days between pre-period fills, year of fill, and plan type (standalone or Medicare Advantage). Results: The proportion of oral anticancer drugs requiring a PA increased from 52.8 to 96.3% from 2010 to 2019. For imatinib, PA increased from 65% to 89% from 2010-16, the year generic imatinib entered. PA for generic imatinib was 95% on market entry. In the imatinib event study sample, there were 1,354 index fills for patients. Unadjusted mean delay in index fill was 40 days for new PA and 15 days for controls; median difference was 4 and 3 days, respectively. Discontinuation was comparable: 9.3% (new PA) vs. 9.5% (control). The adjusted difference between predicted and actual fills for new PA was 20.3 days (95% CI 6.5-34.2, p<0.004). Conclusions: Use of PA for oral oncology drugs has increased substantially over 10 years. In the setting of high prices and provisional outcomes data, PA may be appropriate. However, PA may delay fills, even for high-value drugs like imatinib. Oral anticancer drug adherence is an NCI priority; further study is warranted to identify opportunities to reduce administrative burden of PA and lessen barriers in access to effective medications.

Octet and decuplet baryon σ terms and mass decompositions

We present a comprehensive analysis of the SU(3) octet and decuplet baryon masses and $\sigma$-terms using high-precision lattice QCD data and chiral SU(3) effective theory with finite range regularization. The effects of various systematic uncertainties, including from the scale setting of the lattice data and the regularization prescriptions, are quantified. We find the pion-nucleon and strange nucleon $\sigma$-terms to be $\sigma_{\pi N} = 44(3)(3)$ MeV and $\sigma_{N s} = 50(6)(1)$ MeV, respectively. The results provide constraints on the energy-momentum tensor mass decompositions of the SU(3) octet and decuplet baryons, where we find the trace anomaly and quark/gluon energies decrease for strange baryons due to their larger strange $\sigma$-terms.

Curcuminoids as Anticancer Drugs: Pleiotropic Effects, Potential for Metabolic Reprogramming and Prospects for the Future.

The number of published studies on curcuminoids in cancer research, including its lead molecule curcumin and synthetic analogs, has been increasing substantially during the past two decades. Insights on the diversity of inhibitory effects they have produced on a multitude of pathways involved in carcinogenesis and tumor progression have been provided. As this wealth of data was obtained in settings of various experimental and clinical data, this review first aimed at presenting a chronology of discoveries and an update on their complex in vivo effects. Secondly, there are many interesting questions linked to their pleiotropic effects. One of them, a growing research topic, relates to their ability to modulate metabolic reprogramming. This review will also cover the use of curcuminoids as chemosensitizing molecules that can be combined with several anticancer drugs to reverse the phenomenon of multidrug resistance. Finally, current investigations in these three complementary research fields raise several important questions that will be put among the prospects for the future research related to the importance of these molecules in cancer research.

A model robust subsampling approach for Generalised Linear Models in big data settings

In today’s modern era of big data, computationally efficient and scalable methods are needed to support timely insights and informed decision making. One such method is subsampling, where a subset of the big data is analysed and used as the basis for inference rather than considering the whole data set. A key question when applying subsampling approaches is how to select an informative subset based on the questions being asked of the data. A recent approach for this has been proposed based on determining subsampling probabilities for each data point, but a limitation of this approach is that the appropriate subsampling probabilities rely on an assumed model for the big data. In this article, to overcome this limitation, we propose a model robust approach where a set of models is considered, and the subsampling probabilities are evaluated based on the weighted average of probabilities that would be obtained if each model was considered singularly. Theoretical results are derived to inform such an approach. Our model robust subsampling approach is applied in a simulation study and in two real-world applications where performance is compared to current subsampling practices. The results show that our model robust approach outperforms alternative methods.

Interpretable machine learning for psychological research: Opportunities and pitfalls.

In recent years, machine learning methods have become increasingly popular prediction methods in psychology. At the same time, psychological researchers are typically not only interested in making predictions about the dependent variable, but also in learning which predictor variables are relevant, how they influence the dependent variable, and which predictors interact with each other. However, most machine learning methods are not directly interpretable. Interpretation techniques that support researchers in describing how the machine learning technique came to its prediction may be a means to this end. We present a variety of interpretation techniques and illustrate the opportunities they provide for interpreting the results of two widely used black box machine learning methods that serve as our examples: random forests and neural networks. At the same time, we illustrate potential pitfalls and risks of misinterpretation that may occur in certain data settings. We show in which way correlated predictors impact interpretations with regard to the relevance or shape of predictor effects and in which situations interaction effects may or may not be detected. We use simulated didactic examples throughout the article, as well as an empirical data set for illustrating an approach to objectify the interpretation of visualizations. We conclude that, when critically reflected, interpretable machine learning techniques may provide useful tools when describing complex psychological relationships. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

The presence of intratumoral Porphyromonas gingivalis correlates with a previously defined pancreatic adenocarcinoma, immune cell expression phenotype and with tumor resident, adaptive immune receptor features.

The association between pancreatic adenocarcinoma (PAAD) and the pancreatic microbiome is not fully understood, although bacteria may decrease the effectiveness of chemotherapy and lead to anti-apoptotic, pro-inflammatory microenvironments. To better understand the relationship between the PAAD microbiome and the microenvironment, we identified Porphyromonas gingivalis-positive PAAD samples and found a strong association between intratumoral P. gingivalis and: (i) an immune cell gene expression phenotype previously defined by others as gene program 7; and (ii) recovery of immunoglobulin recombination, sequencing reads. We applied a novel chemical complementarity scoring algorithm, suitable for a big data setting, and determined that the previously established P. gingivalis antigen, rpgB had a reduced chemical complementarity with T-cell receptor (TCR) complementarity-determining region-3 amino acid sequences recovered from PAAD samples with P. gingivalis in comparison to TCR-rpgB chemical complementarity represented by the PAAD samples that lacked P. gingivalis. This finding strengthens the existing body of evidence correlating P. gingivalis with PAAD, which may have implications for the treatment and prognosis of patients. Furthermore, demonstrating the correlation of P. gingivalis and gene program 7 raises the question of whether P. gingivalis infection is responsible for the gene program 7 subdivision of PAAD?

Statistical analysis of high-dimensional biomedical data: a gentle introduction to analytical goals, common approaches and challenges

BackgroundIn high-dimensional data (HDD) settings, the number of variables associated with each observation is very large. Prominent examples of HDD in biomedical research include omics data with a large number of variables such as many measurements across the genome, proteome, or metabolome, as well as electronic health records data that have large numbers of variables recorded for each patient. The statistical analysis of such data requires knowledge and experience, sometimes of complex methods adapted to the respective research questions.MethodsAdvances in statistical methodology and machine learning methods offer new opportunities for innovative analyses of HDD, but at the same time require a deeper understanding of some fundamental statistical concepts. Topic group TG9 “High-dimensional data” of the STRATOS (STRengthening Analytical Thinking for Observational Studies) initiative provides guidance for the analysis of observational studies, addressing particular statistical challenges and opportunities for the analysis of studies involving HDD. In this overview, we discuss key aspects of HDD analysis to provide a gentle introduction for non-statisticians and for classically trained statisticians with little experience specific to HDD.ResultsThe paper is organized with respect to subtopics that are most relevant for the analysis of HDD, in particular initial data analysis, exploratory data analysis, multiple testing, and prediction. For each subtopic, main analytical goals in HDD settings are outlined. For each of these goals, basic explanations for some commonly used analysis methods are provided. Situations are identified where traditional statistical methods cannot, or should not, be used in the HDD setting, or where adequate analytic tools are still lacking. Many key references are provided.ConclusionsThis review aims to provide a solid statistical foundation for researchers, including statisticians and non-statisticians, who are new to research with HDD or simply want to better evaluate and understand the results of HDD analyses.

Study protocol for the use of time series forecasting and risk analyses to investigate the effect of the COVID-19 pandemic on hospital admissions associated with new-onset disability and frailty in a national, linked electronic health data setting

IntroductionOlder people were at particular risk of morbidity and mortality during COVID-19. Consequently, they experienced formal (externally imposed) and informal (self-imposed) periods of social isolation and quarantine. This is hypothesised...